JP7117640B2 - Search method, search device and program - Google Patents

Description

The present disclosure relates to a search method, search device, and program.

Opportunities for users to search for information on products, services, etc. using search services on the Internet are increasing. For example, when searching for candidates for travel destinations or restaurants to be visited by a specific group, it is necessary to input a search keyword that satisfies common tastes of all users belonging to the specific group into an information terminal or the like. .

In Patent Document 1, all users belonging to a specific group input scores for products or services in advance, and based on an evaluation value calculated from the input score history, Techniques for recommending products, services, or the like are disclosed.

JP 2015-18453 A

G. Linden, B. Smith, and J. York, "Amazon.com Recommendations: Item-to-Item Collaborative Filtering," IEEE Internet Computing, vol. 7, no. 1, 2003, pp. 76-80.

However, with the technology disclosed in the above-mentioned Patent Document 1, the user needs to input the score in advance. It is not possible to recommend products, services, etc. for which is not entered in advance to a specific group.

In addition, the technique disclosed in Non-Patent Document 1 does not require the user to input the score in advance, but is based on actual purchasing behavior, so it is not applicable to product fields where there is no information on purchasing behavior. could not.

Therefore, the present disclosure provides a search method, search device, and program capable of recommending optimal products or services to a specific group without using purchase histories.

A search method according to an aspect of the present disclosure is a search method executed by a processor, the processor: (a) obtaining a search word; (b) a plurality of words and a first concept map, a second concept map and a third concept map comprising a semantic distance between the first concept map, the first concept map being specific to the first user belonging to the first group; storing in memory a second concept map specific to a second user belonging to the first group and a third concept map specific to a third user belonging to a second group different from the first group; and (c) a first degree of relevance including a degree of relevance indicating the closeness between the semantic distance contained in the first concept map and the semantic distance contained in the second concept map (d) determining the semantic distances contained in the first concept map, the semantic distances contained in the second concept map, and the semantic distances contained in the third concept map; obtaining a second relevance map including relevance indicating distance and closeness; , extracting as a related word at least one word whose difference from the relevance to the search word contained in the second relevance map is equal to or greater than a first threshold; (f) the search word and the related word; Output search results based on .

In addition, these general or specific aspects may be realized by an apparatus, system, integrated circuit, computer program, or computer-readable recording medium. Any combination of media may be implemented. Computer-readable recording media include non-volatile recording media such as CD-ROMs (Compact Disc-Read Only Memory).

According to a search method or the like according to an aspect of the present disclosure, it is possible to recommend an optimal product, service, or the like to a specific group without using purchase histories. Additional benefits and advantages of aspects of the disclosure will be apparent from the specification and drawings. This benefit and/or advantage may be provided individually by the various aspects and features disclosed in the specification and drawings, and not all are required to obtain one or more thereof.

1 is a diagram showing an application example of a search system according to Embodiment 1; FIG. FIG. 10 is a diagram showing another application example of the search system according to Embodiment 1; 1 is a block diagram showing an overview of a search system according to Embodiment 1; FIG. 1 is a block diagram showing the configuration of a search system according to Embodiment 1; FIG. It is a figure which shows an example of a 1st concept map. It is a figure which shows an example of a 2nd concept map. It is a figure which shows an example of a 3rd concept map. It is a figure which shows an example of a 4th concept map. FIG. 2 is a diagram for conceptually explaining first to fourth concept maps; FIG. FIG. 2 is a diagram for conceptually explaining first to fourth concept maps; FIG. 4 is a flow chart showing the flow of operations of a group common concept extraction unit according to Embodiment 1; FIG. 8 is a diagram showing an example of calculation of inter-concept distances by a group-common concept extraction unit according to Embodiment 1; FIG. 7 is a diagram showing an example of calculation of a degree of association by a group common concept extraction unit according to Embodiment 1; 5 is a flow chart showing the flow of operations of a global common concept extraction unit according to Embodiment 1; FIG. 10 is a diagram showing an example of calculation of inter-concept distances by the globally common concept extraction unit according to Embodiment 1; FIG. 10 is a diagram showing an example of calculation of a degree of association by the overall common concept extraction unit according to Embodiment 1; 4 is a flow chart showing the flow of operations of a global common concept exclusion unit according to Embodiment 1. FIG. FIG. 4 schematically shows a correlation matrix extracted from a first concept map; FIG. 4 is a diagram schematically showing correlation matrices respectively extracted from a first conceptual map and a second conceptual map; FIG. 4 is a diagram schematically showing correlation matrices extracted from first to fourth concept maps; FIG. 5 is a diagram showing an operation example of a recommendation requesting unit according to Embodiment 1; FIG. 8 is a diagram showing a display example 1 of a recommendation result on the display unit of the recommendation target user terminal according to Embodiment 1; FIG. 8 is a diagram showing a display example 1 of a recommendation result on the display unit of the recommendation target user terminal according to Embodiment 1; FIG. 10 is a diagram showing a display example 2 of a recommendation result on the display unit of the recommendation target user terminal according to Embodiment 1; FIG. 10 is a diagram showing a display example 3 of a recommendation result on the display unit of the recommendation target user terminal according to Embodiment 1; 2 is a block diagram showing the configuration of a search system according to Embodiment 2; FIG. FIG. 11 is a diagram showing an example of calculation of conceptual average distances by a conceptual average distance calculation unit according to Embodiment 2; 10 is a flow chart showing the flow of operations of a global common concept exclusion unit according to Embodiment 2. FIG. FIG. 11 is a block diagram showing the configuration of a search system according to Embodiment 3; FIG. 13 is a block diagram showing the configuration of a search system according to Embodiment 4; FIG.

A search method according to an aspect of the present disclosure is a search method executed by a processor, the processor: (a) obtaining a search word; (b) a plurality of words and a first concept map, a second concept map and a third concept map comprising a semantic distance between the first concept map, the first concept map being specific to the first user belonging to the first group; storing in memory a second concept map specific to a second user belonging to the first group and a third concept map specific to a third user belonging to a second group different from the first group; and (c) a first degree of relevance including a degree of relevance indicating the closeness between the semantic distance contained in the first concept map and the semantic distance contained in the second concept map (d) determining the semantic distances contained in the first concept map, the semantic distances contained in the second concept map, and the semantic distances contained in the third concept map; obtaining a second relevance map including relevance indicating distance and closeness; , extracting as a related word at least one word whose difference from the relevance to the search word contained in the second relevance map is equal to or greater than a first threshold; (f) the search word and the related word; Output search results based on .

According to this aspect, based on the first concept map and the second concept map, related words indicating a concept common to the search word for the first user and the second user belonging to the first group are extracted. . Since a search is performed based on the extracted related words and search words, even if the first user or the second user has never purchased a product or used a service, etc., the first group can It is possible to recommend the most suitable product or service for the customer. Further, the difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map out of the plurality of words is equal to or greater than the first threshold. Extract words as related words. As a result, it is possible to extract related words indicating concepts common only to the first user and the second user belonging to the first group, and to increase the accuracy of extracting related words.

For example, in (e), for the at least one word whose difference is equal to or greater than the first threshold, the semantic distance from the search word included in the first concept map and the second concept If a value based on the sum of the semantic distances to the search words included in the map is equal to or less than a second threshold, the search words may be extracted as the related words.

According to this aspect, it is possible to exclude words that are semantically distant from the search word from the related words, although the concept is common to the search word for the first user and the second user. This makes it possible to recommend more optimal products or services to the first group.

For example, in (f) above, a search result based on the search word and the related word is displayed to the first user or the second user belonging to the first group. may

According to this aspect, by displaying the search results, it is possible to recommend products, services, or the like to the first group.

For example, in (f) above, the search result based on the search word and the related word is displayed in descending order of the relevance between the search word and the related word included in the first relevance map. It may be configured as

According to this aspect, the search results can be displayed in order of recommendation for the first group.

For example, in (f) above, the search result based on the search word and the related word may be displayed at a higher rank than the content highly evaluated by other users.

According to this aspect, search results to be recommended to the first group can be preferentially displayed.

For example, each of the first conceptual map, the second conceptual map, and the third conceptual map may be generated based on brain measurement results.

According to this aspect, the first conceptual map, the second conceptual map, and the third conceptual map can be generated based on the result of brain measurement.

For example, each of the first conceptual map, the second conceptual map, and the third conceptual map is configured to be generated based on brain measurement results using fMRI (Functional Magnetic Resonance Imaging). may

According to this aspect, the first conceptual map, the second conceptual map, and the third conceptual map can be generated based on the results of brain measurement using fMRI.

A search device according to an aspect of the present disclosure is a search device comprising a processor and a memory, wherein the processor (a) obtains a search word; (b) a plurality of words; a first concept map, a second concept map, and a third concept map, wherein the first concept map is specific to the first user belonging to the first group, and a semantic distance between each a second concept map specific to a second user belonging to the first group; and a third concept map specific to a third user belonging to a second group different from the first group; from the memory, and (c) a first (d) the semantic distances contained in the first concept map, the semantic distances contained in the second concept map, and the semantic distances contained in the third concept map obtaining a second relevance map including relevance indicating closeness to the semantic distance; (f) extracting at least one word having a difference between the degree of relevance and the degree of relevance to the search word contained in the second degree-of-relevance map equal to or greater than a first threshold as a related word; A search word is output based on the related word.

According to this aspect, based on the first concept map and the second concept map, related words indicating a concept common to the search word for the first user and the second user belonging to the first group are extracted. . Since a search is performed based on the extracted related words and search words, even if the first user or the second user has never purchased a product or used a service, etc., the first group can It is possible to recommend the most suitable product or service for the customer. Further, the difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map out of the plurality of words is equal to or greater than the first threshold. Extract words as related words. As a result, it is possible to extract related words indicating concepts common only to the first user and the second user belonging to the first group, and to increase the accuracy of extracting related words.

For example, in (e), for the at least one word whose difference is equal to or greater than the first threshold, the semantic distance from the search word included in the first concept map and the second concept If a value based on the sum of the semantic distances to the search words included in the map is equal to or less than a second threshold, the search words may be extracted as the related words.

According to this aspect, it is possible to exclude words that are semantically distant from the search word from the related words, although the concept is common to the search word for the first user and the second user. This makes it possible to recommend more optimal products or services to the first group.

For example, the processor may further: (g) display search results based on the search word and the related word to the first user or the second user belonging to the first group; can be configured to

According to this aspect, by displaying the search results, it is possible to recommend products, services, or the like to the first group.

For example, in (g) above, the search result based on the search word and the related word is displayed in descending order of the relevance between the search word and the related word included in the first relevance map. It may be configured as

According to this aspect, the search results can be displayed in order of recommendation for the first group.

For example, in (g) above, the search result based on the search word and the related word may be displayed at a higher rank than the content highly evaluated by other users.

According to this aspect, search results to be recommended to the first group can be preferentially displayed.

For example, each of the first conceptual map, the second conceptual map, and the third conceptual map may be generated based on brain measurement results.

According to this aspect, the first conceptual map, the second conceptual map, and the third conceptual map can be generated based on the result of brain measurement.

For example, each of the first conceptual map, the second conceptual map, and the third conceptual map may be generated based on results of brain measurement using fMRI.

According to this aspect, the first conceptual map, the second conceptual map, and the third conceptual map can be generated based on the results of brain measurement using fMRI.

A program according to an aspect of the present disclosure is a program that causes a computer to execute a search method, the computer: (a) obtaining a search word; (b) a plurality of words; a first concept map, a second concept map and a third concept map comprising a semantic distance between the first concept map, the first concept map being specific to the first user belonging to the first group; storing in memory a second concept map specific to a second user belonging to the first group and a third concept map specific to a third user belonging to a second group different from the first group; and (c) a first degree of relevance including a degree of relevance indicating the closeness between the semantic distance contained in the first concept map and the semantic distance contained in the second concept map (d) determining the semantic distances contained in the first concept map, the semantic distances contained in the second concept map, and the semantic distances contained in the third concept map; obtaining a second relevance map including relevance indicating distance and closeness; , extracting as a related word at least one word whose difference from the relevance to the search word contained in the second relevance map is equal to or greater than a first threshold; (f) the search word and the related word; to output the search results based on

According to this aspect, based on the first concept map and the second concept map, related words indicating a concept common to the search word for the first user and the second user belonging to the first group are extracted. . Since a search is performed based on the extracted related words and search words, even if the first user or the second user has never purchased a product or used a service, etc., the first group can It is possible to recommend the most suitable product or service for the customer. Further, the difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map out of the plurality of words is equal to or greater than the first threshold. Extract words as related words. As a result, it is possible to extract related words indicating concepts common only to the first user and the second user belonging to the first group, and to increase the accuracy of extracting related words.

In addition, these general or specific aspects may be realized by a system, method, integrated circuit, computer program, or a recording medium such as a computer-readable CD-ROM. Any combination of programs or recording media may be used.

Hereinafter, embodiments will be specifically described with reference to the drawings.

It should be noted that the embodiments described below are all comprehensive or specific examples. Numerical values, shapes, materials, components, arrangement positions and connection forms of components, steps, order of steps, and the like shown in the following embodiments are examples, and are not intended to limit the present disclosure. In addition, among the constituent elements in the following embodiments, constituent elements that are not described in independent claims representing the highest concept will be described as arbitrary constituent elements.

(Embodiment 1)
[1-1. Overview of search system]
First, an outline of a search system 2 according to Embodiment 1 will be described with reference to FIGS. 1A to 2. FIG. FIG. 1A is a diagram showing an application example of the search system 2 according to the first embodiment. FIG. 1B is a diagram showing another application example of the search system 2 according to the first embodiment. FIG. 2 is a block diagram showing an overview of the search system 2 according to Embodiment 1. As shown in FIG.

As shown in FIGS. 1A and 1B, the search system 2 according to Embodiment 1 executes a search on an EC (Electronic Commerce) site or the like when a user 6 belonging to a specific group 4 inputs a search word. It is a system for outputting the search results. As a result, the search system 2 recommends products, services, or the like that satisfy common preferences of all the users 6 belonging to the specific group 4 to the specific group 4 .

In the example shown in FIG. 1A, the search system 2 comprises a voice terminal 8 having a voice interface. The user 6 inputs a search word to the voice terminal 8 by speaking to the voice terminal 8 . As a result, the voice terminal 8 outputs the search result by voice. Specifically, when the user 6 inputs a search word for requesting a recommended restaurant or the like into the voice terminal 8, the voice terminal 8 provides an appropriate service corresponding to the search word by voice. .

In the example shown in FIG. 1B, the search system 2 includes an information terminal 10 such as a tablet or smart phone having a touch panel. The user 6 inputs search words by operating the touch panel of the information terminal 10 . Thereby, the information terminal 10 displays the search result on the touch panel.

As shown in FIG. 2, the search system 2 includes a recommended concept calculator 12 (an example of a search device), a recommended service provider 14, a plurality of recommended user terminals 16a and 16b, and a plurality of non-recommended user terminals. 18a and 18b. These are connected via the Internet 20 so as to be able to communicate with each other.

A plurality of recommended user terminals 16a and 16b are information terminals operated by a first user and a second user belonging to a first group, respectively. The first group is a group to which the search system 2 recommends products, services, or the like. The plurality of recommended user terminals 16 a and 16 b each receive search words input by the first user and the second user, and transmit the received search words to the recommended concept calculation unit 12 .

The plurality of non-recommendable user terminals 18a and 18b are information terminals operated by a third user and a fourth user who belong to a second group different from the first group, respectively. The second group is a group to which the search system 2 does not recommend products, services, or the like.

The recommended concept calculation unit 12 receives search words transmitted from each of the plurality of recommended user terminals 16a and 16b. The recommended concept calculation unit 12 adds, to the received search words, related words indicating group common concepts common to the first user and the second user belonging to the first group. A search word that is a combination with a word is transmitted to the recommendation service providing unit 14 as a recommended concept.

The recommendation service providing unit 14 receives the recommended concepts transmitted from the recommended concept calculation unit 12 . The recommendation service providing unit 14 performs a search based on the received recommendation concept, and transmits the search result as a recommendation result to at least one of the plurality of recommended user terminals 16a and 16b.

[1-2. Configuration of recommended user terminal]
Next, the configuration of the recommended user terminal 16a will be described with reference to FIG. FIG. 3 is a block diagram showing the configuration of the search system 2 according to Embodiment 1. As shown in FIG. Since the plurality of recommended user terminals 16a and 16b have the same configuration, only the configuration of the recommended user terminal 16a will be described below.

As shown in FIG. 3, the recommendation target user terminal 16a has a search word transmission unit 22, a concept map holding unit 24, and a recommendation result reception unit .

The search word transmission unit 22 transmits the search words input by the first user who is the user of the recommendation target user terminal 16 a to the recommended concept calculation unit 12 .

The concept map holding unit 24 is a memory that holds a first concept map unique to the first user. The concept map holding unit 24 transmits the first concept map to the recommended concept calculation unit 12 based on the request from the recommended concept calculation unit 12 .

The recommendation result receiving unit 26 receives the recommendation result transmitted from the recommendation service providing unit 14 . The recommendation result received by the recommendation result receiving unit 26 is displayed, for example, on a display unit 70 (see FIG. 16A described later) such as a touch panel mounted on the recommendation target user terminal 16a.

The concept map holding unit of the recommendation target user terminal 16b holds a second concept map unique to the second user who is the user of the recommendation target user terminal 16b. Further, each of the plurality of non-recommendable user terminals 18a and 18b has a concept map holding unit as described above. The concept map holding unit of the non-recommendable user terminal 18a holds a third concept map unique to the third user who is the user of the non-recommendable user terminal 18a. The concept map holding unit of the non-recommendable user terminal 18b holds a fourth concept map unique to the fourth user who is the user of the non-recommendable user terminal 18b.

Here, each configuration of the above-described first to fourth conceptual maps will be described with reference to FIGS. 4A to 6. FIG. FIG. 4A is a diagram showing an example of a first concept map. FIG. 4B is a diagram showing an example of a second concept map. FIG. 4C is a diagram showing an example of a third concept map. FIG. 4D is a diagram showing an example of a fourth concept map. 5 and 6 are diagrams for conceptually explaining the first to fourth conceptual maps.

The first to fourth concept maps are associated with the first to fourth users, respectively, and are expressed as two-dimensional tables shown in FIGS. 4A to 4D, for example. As shown in FIGS. 4A to 4D, each of the first to fourth concept maps includes multiple general words (one example of words) and semantic distances between each of the multiple general words. .

Each of the first to fourth concept maps contains, for example, 1000 kinds of general words. In this case, each of the first to fourth concept maps is a two-dimensional table with a size of 1000 rows×1000 columns. Each of the plurality of general words is an English word indicating a concept used in daily life, such as "home", "rent", "owner" and "house". For convenience of explanation, in FIGS. 4A to 4D, each of the first to fourth concept maps is represented as a two-dimensional table of 8 rows×8 columns size containing 8 types of words.

The semantic distance is a value expressing the closeness of the semantic distance between each of a plurality of general words in relative magnitude. In the examples shown in FIGS. 4A-4D, the semantic distance is represented by an integer value from '0' to '10'. A smaller semantic distance value represents a closer semantic distance between each of the plurality of general words. Specifically, when the semantic distance is "1", the semantic distance between two different general words is the closest, and when the semantic distance is "10", two different The semantic distance between common words of the class is the furthest. It is assumed that the semantic distance between common words of the same type, such as between "home" and "home", is "0".

The semantic distance between the first user's 'first word' and 'second word' indicates how easily the user associates the 'first word' and the 'second word'. It may be information.

In the example shown in FIG. 4A, the semantic distance to "home" is "1" for "house", "car" and "weekend", "2" for "rent" and "place", and "owner ' is '4' and 'park' is '5'. On the other hand, in the example shown in FIG. 4B, the semantic distance to "home" is "1" for "house" and "car", "2" for "owner" and "place", and "2" for "rent". "3" and "park" and "weekend" are "5".

For the first user, for example, "home" and "weekend" are semantically close concepts, and it is considered that the first user often spends weekends at home. On the other hand, for the second user, for example, "home" and "weekend" are semantically distant concepts, and it is considered that the second user often goes out on weekends. That is, the difference in semantic distance between the first conceptual map shown in FIG. 4A and the second conceptual map shown in FIG. represents. Similarly, the difference in semantic distance between the third concept map shown in FIG. 4C and the fourth concept map shown in FIG. represents.

Each of the first to fourth concept maps described above is generated based on, for example, brain measurement results. Specifically, when the 1st to 4th users are listening to a plurality of stories by voice, the brain reactions of the 1st to 4th users are analyzed by fMRI (functional Magnetic Resonance Imaging). measure. The stories contain about 1000 English words. Based on the measurement results of brain responses, the relationships between each of a plurality of English words are visually mapped on the cerebral cortex. For example, the semantic distance between each of 1000 English words can be defined by the method described in the following document. That is, a concept map can be generated having semantic distances between each of a plurality of English words, as shown in FIGS. 5 and 6, for example. 5 and 6 are concept maps specific to different users, and the physical distance between each of the English words in FIGS. 5 and 6 represents individual differences in concept between users. there is For example, the physical distance between each of a plurality of English words on the cerebral cortex can be normalized to define a semantic distance ranging from '0.0' to '1.0'.

For the brain measurement described above, see, for example, "Alexander G. Huth, Wendy A. de Heer, Thomas L. Griffiths, Frederic E. Theunissen, Jack L. Gallant, "Natural speech reveals the semantic maps that tile brain". Nature, Vol.532, p.453-458, Apr. 28, 2016, Nature Publishing Group”, and other known methods can be used.

Note that the concept map may be generated by other methods other than the method of generating based on the results of brain measurement described above, as long as the semantic distances between each of the plurality of concepts are defined. For example, the concept map may be generated by estimating the semantic distance between each of the plurality of concepts based on the first to fourth users' remarks and actions in their daily lives.

[1-3. Configuration of Recommended Concept Calculation Unit]
Next, the configuration of the recommended concept calculation unit 12 will be described with reference to FIG. As shown in FIG. 3, the recommended concept calculation unit 12 includes a search word reception unit 28, a recommended terminal concept map reception unit 30, a non-recommended terminal concept map reception unit 32, a group common concept extraction unit 34, It has a global common concept extraction unit 36 , a global common concept exclusion unit 38 , and a recommended concept transmission unit 40 .

The search word receiving unit 28 receives search words transmitted from each of the search word transmitting unit 22 of the recommended user terminal 16a and the search word transmitting unit of the recommended user terminal 16b.

The recommendation target terminal concept map reception unit 30 receives the first concept map and the second concept map respectively transmitted from the concept map holding unit 24 of the recommendation target user terminal 16a and the concept map holding unit of the recommendation target user terminal 16b. do.

The non-recommendable terminal concept map receiving unit 32 receives the third conceptual map and the fourth conceptual map transmitted from the conceptual map holding unit of the non-recommended user terminal 18a and the conceptual map holding unit of the non-recommended user terminal 18b, respectively. receive.

Based on the first concept map and the second concept map received by the recommendation target terminal concept map receiving unit 30, the group common concept extraction unit 34 extracts at least one common concept map to the first concept map and the second concept map. One general word is extracted as a group common concept.

The overall common concept extraction unit 36 extracts the first concept map and the second concept map received by the recommendation target terminal concept map reception unit 30 and the third concept received by the non-recommendation target terminal concept map reception unit 32. Based on the map and the fourth concept map, extracting at least one general word common to the first concept map, the second concept map, the third concept map and the fourth concept map as an overall common concept .

The overall common concept exclusion unit 38 excludes the overall common concept extracted by the overall common concept extraction unit 36 from the group common concept extracted by the group common concept extraction unit 34, thereby excluding the overall common concept received by the search word receiving unit 28. A related word highly related to the search word and indicating a concept common only to the first group is extracted. The global common concept exclusion unit 38 outputs the combination of the search word and the related word to the recommended concept transmission unit 40 as a recommended concept.

The recommended concept transmitting unit 40 transmits the recommended concepts output from the universal concept excluding unit 38 to the recommended service providing unit 14 .

[1-4. Configuration of Recommendation Service Provider]
Next, the configuration of the recommendation service providing unit 14 will be described with reference to FIG. As shown in FIG. 3 , the recommendation service providing unit 14 has a recommended concept receiving unit 42 , a database 44 , a recommendation requesting unit 46 , a recommendation content generating unit 48 , and a recommendation result transmitting unit 50 .

The recommended concept reception unit 42 receives the recommended concept transmitted from the recommended concept calculation unit 12 and outputs the received recommended concept to the recommendation request unit 46 .

The database 44 stores data relating to, for example, residential properties for recommendation to the first group or the like.

The recommendation requesting unit 46 refers to the database 44 to extract data most relevant to the recommended concept output from the recommended concept receiving unit 42 .

The recommendation content generation unit 48 converts the data extracted by the recommendation request unit 46 into a format suitable for display on the display unit 70 (see FIG. 16A described later) of the recommendation target user terminal 16a. The recommendation content generation unit 48 outputs the converted data to the recommendation result transmission unit 50 as recommendation content for recommending to the first group.

The recommendation result transmission unit 50 transmits the recommendation content output from the recommendation content generation unit 48 to the recommendation target user terminal 16a to which the search word is input, for example, as a recommendation result.

[1-5. Operation of search system]
[1-5-1. Operation of Group Common Concept Extraction Unit]
The operation of the group common concept extraction unit 34 will be described with reference to FIGS. 7 to 8B. FIG. 7 is a flow chart showing the flow of operations of the group common concept extraction unit 34 according to the first embodiment. FIG. 8A is a diagram showing an example of calculation of inter-concept distances by the group-common concept extraction unit 34 according to the first embodiment. FIG. 8B is a diagram showing an example of calculation of the degree of association by the group common concept extraction unit 34 according to the first embodiment.

A case will be described below in which the first user belonging to the first group inputs the search word "home" into the recommendation target user terminal 16a in order to search for housing properties suitable for the first group, for example.

As shown in FIG. 7, first, the group common concept extraction unit 34 acquires the first concept map and the second concept map from the recommendation target terminal concept map reception unit 30 (S101). Next, the group common concept extraction unit 34 calculates the inter-concept distance between the first concept map and the second concept map based on the following equation 1 (S102).

In Equation 1 above, D _ij is the inter-concept distance at i row and j column, d _Aij is the semantic distance at i row and j column of the first concept map, and d _Bij is the i row of the second concept map. It is the semantic distance of the j column. For example, the semantic distance of i row j column of the first concept map shown in FIG. is the semantic distance (e.g., "2") between a common word (e.g., "rent").

Here, the inter-concept distance represents the closeness of the semantic distance between the first concept map and the second concept map. The smaller the inter-concept distance, the semantic distance between the general word in row i and the general word in column j of the first concept map, and the general word in row i and the general word in column j of the second concept map. It means that the semantic distance between and is close. As shown in FIG. 8A, the group common concept extraction unit 34 calculates the inter-concept distance between the first concept map shown in FIG. 4A and the second concept map shown in FIG. 4B based on the above equation 1. do.

In this embodiment, the case where two users belong to the first group has been described. An inter-concept distance may be calculated for any two concept maps extracted from one or more concept maps.

Next, the group common concept extraction unit 34 converts the inter-concept distance D _ij calculated in step S102 into a degree of association v _ij based on the following equation 2 (S103).

Here, the degree of association represents the closeness of the semantic distance between the first concept map and the second concept map. For example, the group common concept extraction unit 34 acquires the first relevance map shown in FIG. 8B by converting the inter-concept distances shown in FIG. 8A into relevance based on Equation 2 above. The first relevance map is a two-dimensional table in which the inter-concept distances between the first concept map and the second concept map are converted into relevance.

In the example shown in FIG. 8B, the degree of association is a positive value of "1.00" or less. The greater the degree of relevance, the greater the semantic distance between the general word in row i and the general word in column j of the first concept map, and the general word in row i and the general word in column j of the second concept map. It means that the semantic distance between is close. For example, the relevance of 'car' to 'home' is '1.00', which means that 'home' and 'car' are the most relevant for the first user and the second user. represents. In FIG. 8B, the row surrounded by the dashed frame line 52 represents the degree of relevance of each of a plurality of general words with respect to the search word "home" in the first degree of relevance map.

In this embodiment, the group common concept extraction unit 34 extracts all general words shown in FIG. 8B as group common concepts. Note that the group common concept extraction unit 34 may extract general words having a degree of association equal to or higher than a predetermined threshold value (for example, "0.50") as group common concepts.

Finally, the group common concept extraction unit 34 outputs the first relevance map acquired in step S103 to the overall common concept exclusion unit 38 (S104).

[1-5-2. Operation of Global Common Concept Extraction Unit]
The operation of the general common concept extraction unit 36 will be described with reference to FIGS. 9 to 10B. FIG. 9 is a flow chart showing the operation flow of the global common concept extraction unit 36 according to the first embodiment. FIG. 10A is a diagram showing an example of calculation of inter-concept distances by the global common concept extraction unit 36 according to the first embodiment. FIG. 10B is a diagram showing an example of calculation of the degree of association by the overall common concept extraction unit 36 according to Embodiment 1. As shown in FIG.

As shown in FIG. 9, first, the overall common concept extraction unit 36 acquires the first concept map and the second concept map from the recommendation target terminal concept map reception unit 30 (S201). Next, the overall common concept extraction unit 36 acquires the third concept map and the fourth concept map from the non-recommendable terminal concept map reception unit 32 (S202). Next, the overall common concept extraction unit 36 calculates the inter-concept distance between the first to fourth concept maps based on the following Equation 3 (S203).

In the above equation 3, D' _ij is the inter-concept distance at i row and j column, and d _xij and d _yij are respectively i row j Semantic distance of columns. As shown in FIG. 10A, the overall common concept extraction unit 36 calculates the inter-concept distances between the first to fourth concept maps respectively shown in FIGS. 4A to 4D, based on Equation 3 above.

For example, for "car" and "weekend", D' ₇₈ =|d ₁₇₈ -d ₂₇₈ | ² +|d ₁₇₈ -d ₃₇₈ | ² +|d ₁₇₈ -d ₄₇₈ | ² +|d ₂₇₈ -d ₃₇₈ ² +|d ₂₇₈ -d ₄₇₈ | ² ++|d ₃₇₈ -d ₄₇₈ | ² =|2-1| ² +|2-4| ² +|2-1| ² +|1-4| ² + |1-1| ² +|4-1| ² .

Next, the overall common concept extraction unit 36 converts the inter-concept distance D' _ij calculated in step S203 into the degree of association v' _ij based on the following equation 4 (S204).

For example, the global common concept extraction unit 36 acquires the second relevance map shown in FIG. 10B by converting the inter-concept distances shown in FIG. 10A into relevance based on Equation 4 above. The second relevance map is a two-dimensional table in which inter-concept distances between the first to fourth concept maps are converted into relevance. For example, the degree of relevance of 'house' to 'home' is '1.00', which means that 'home' and 'house' have the highest degree of relevance for the first to fourth users. ing. In FIG. 10B, the line surrounded by the dashed frame line 54 represents the degree of relevance of each of the plurality of general words with respect to the search word "home" in the second degree of relevance map.

In the present embodiment, the global common concept extraction unit 36 extracts all general words shown in FIG. 10B as global common concepts. Note that the global common concept extraction unit 36 may extract general words having a degree of association equal to or higher than a predetermined threshold value (for example, "0.50") as the global common concept.

Finally, the global common concept extraction unit 36 outputs the second relevance map acquired in step S204 to the global common concept exclusion unit 38 (S205).

[1-5-3. Operation of Global Common Concept Exclusion Part]
The operation of the global common concept exclusion unit 38 will be described with reference to FIG. FIG. 11 is a flow chart showing the operation flow of the global common concept exclusion unit 38 according to the first embodiment.

As shown in FIG. 11, first, the overall common concept exclusion unit 38 acquires the first relevance map and the second relevance map from the group common concept extraction unit 34 and the overall common concept extraction unit 36, respectively (S301). ). Next, the global common concept exclusion unit 38 acquires the search word "home" from the search word reception unit 28 (S302).

Next, the global common concept exclusion unit 38 acquires the relevance v of each of the plurality of general words with respect to the search word "home" from the first relevance map (S303). Specifically, the global common concept exclusion unit 38 acquires the relevance v of each of the general words in the line surrounded by the dashed frame line 52 from the first relevance map shown in FIG. 8B.

Next, the global common concept exclusion unit 38 acquires the relevance v' of each of the plurality of general words with respect to the search word "home" from the second relevance map (S304). Specifically, the overall common concept exclusion unit 38 acquires the relevance v′ of each of the multiple general words in the line surrounded by the dashed frame 54 from the second relevance map shown in FIG. 10B. .

Next, the global common concept exclusion unit 38 extracts one general word from among a plurality of general words, and calculates the difference v−v′ between the degree of relevance v and the degree of relevance v′ for the extracted general word ( S305). The overall common concept exclusion unit 38 determines whether or not the calculated difference v−v′ is equal to or greater than a first threshold value (for example, “0.20”) (S306).

If the difference v−v′ is greater than or equal to the first threshold (YES in S306), the global common concept exclusion unit 38 adopts the extracted general words as related words (S307). On the other hand, if the difference v−v′ is less than the first threshold (NO in S306), the global common concept exclusion unit 38 does not adopt the extracted general word as a related word (S308). In the examples shown in FIGS. 8B and 10B, "rent" (vv'=0.30), "park" (vv'=0.96) and "car" (vv'=0 .96) are adopted as related words.

If the difference v-v' has not been calculated for all general words (NO in S309), step S305 is executed again. On the other hand, when the difference v−v′ is calculated for all general words (YES in S309), the overall common concept exclusion unit 38 removes the search word “home” and the related words “rent”, “park” and “car”. ” as a recommended concept to the recommended concept transmission unit 40 (S310).

Here, the processing of the global common concept exclusion unit 38 will be conceptually described with reference to FIGS. 12 to 14. FIG. FIG. 12 is a diagram schematically showing a correlation matrix generated from the first concept map. FIG. 13 is a diagram schematically showing correlation matrices respectively extracted from the first conceptual map and the second conceptual map. FIG. 14 is a diagram schematically showing correlation matrices extracted from the first to fourth concept maps.

FIG. 12 illustrates a correlation matrix between the search word "home" and each of a plurality of general words generated from the first concept map. The horizontal axis of FIG. 12 represents a list in which a plurality of general words are arranged in a horizontal line. In FIG. 12, general words having a close semantic distance to the search word "home" are hatched.

FIG. 13 illustrates the correlation matrix generated from the first concept map and the correlation matrix generated from the second concept map. These two correlation matrices are assumed to be generated by the same procedure as the correlation matrix shown in FIG. By comparing the two correlation matrices shown in FIG. 13, common words having a close semantic distance to the search word "home", which are common between the first user and the second user, are extracted. can be done. However, the general words extracted in this manner include not only the group common concept 56 (for example, "rent") that is common only between the first user and the second user, but also the common words of the first to fourth users. The group common concept 56 and the global common concept 58 cannot be distinguished because they may also contain an overall common concept 58 (eg, "house") that is common between them. If the extracted general words include the global common concept 58, there arises a problem that the general words indicating the concept common only to the first user and the second user cannot be adopted as related words.

FIG. 14 shows correlation matrices respectively generated from the first to fourth concept maps. By comparing the four correlation matrices shown in FIG. 14, it is possible to extract general words that are common among the first to fourth users and have a close semantic distance to the search word "home." The general words extracted in this way do not include the overall common concept 60 common among the first to fourth users, and the group common concept 62 common only between the first user and the second user is included. Therefore, a general word indicating a concept common only to the first user and the second user can be adopted as the related word. That is, the problem as shown in FIG. 13 described above can be solved by executing the steps S301 to S310 described above by the global common concept exclusion unit .

[1-5-4. Operation example of the recommendation request unit]
An operation example of the recommendation requesting unit 46 of the recommendation service providing unit 14 will be described with reference to FIG. 15 . FIG. 15 is a diagram showing an operation example of the recommendation requesting unit 46 according to the first embodiment.

The recommendation requesting unit 46 holds information regarding search queries that are most relevant to each of the plurality of general words. As shown in FIG. 15, the recommendation requesting unit 46 selects related words 64 (for example, “rent”, “park” and “car”) included in the recommended concepts output from the recommended concept receiving unit 42 based on the information. ), and query the property database 68 in the database 44 to prioritize the search query 66 that is most relevant. In the example shown in FIG. 15, the search query 66 is the contract terms for "rent", the distance to a nearby park for "park", and the availability of a parking lot for "car".

As a result of the inquiry to the property database 68, the recommendation requesting unit 46 selects “property E” from among “property A” to “property F” stored in the property database 68, for example, as the data most relevant to the recommended concept. Extract as

[1-5-5. Display example of recommendation results]
Next, display examples of recommendation results on the display unit 70 of the recommendation target user terminal 16a will be described with reference to FIGS. 16A to 18. FIG. 16A and 16B are diagrams showing a display example 1 of recommendation results on the display unit 70 of the recommendation target user terminal 16a according to the first embodiment. FIG. 17 is a diagram showing a display example 2 of a recommendation result on the display unit 70 of the recommendation target user terminal 16a according to the first embodiment. FIG. 18 is a diagram showing a display example 3 of a recommendation result on the display unit 70 of the recommendation target user terminal 16a according to the first embodiment.

In the display example 1 shown in FIG. 16A, the display unit 70 of the recommendation target user terminal 16a has a display area 72 for displaying an input search word (for example, "home") and a related word included in the recommendation result ( For example, a display area 74 for displaying "rent", "park" and "car") and option buttons 76 and 78 are displayed. The first user touches the option button 76 if he wishes to display the recommendation results, and touches the option button 78 if he does not wish to display the recommendation results.

When the first user touches the option button 76, a recommendation result display screen 80 is displayed on the display unit 70 of the recommendation target user terminal 16a, as shown in FIG. 16B. Each recommendation result from the first recommendation result 82 to the Nth recommendation result 84 is displayed on the recommendation result display screen 80 . For example, the first recommendation result 82 includes the photograph and description of the first recommended property. Note that the display order of the plurality of recommendation results on the recommendation result display screen 80 may be the descending order of the degree of association included in the first degree-of-association map.

In display example 2 shown in FIG. 17, the display unit 70 of the recommendation target user terminal 16a has a display area 72 for displaying the input search word and a display area for displaying related words included in the recommendation result. 74 and a display area 86 for displaying recommendation results. In the display area 86, each recommendation result from the first recommendation result 82 to the Nth recommendation result 84 is displayed.

In the display example 3 shown in FIG. 18, the display unit 70 of the recommendation target user terminal 16a has a display area 72 for displaying input search words and a display area for displaying related words included in the recommendation results. 74, a display area 86 for displaying recommendation results, and a display area 88 for displaying properties highly rated by general users. The display area 86 is displayed higher than the display area 88 . Note that the display area 88 displays content that is highly popular among the plurality of users 6 who use the search system 2 .

[1-6. effect]
As described above, based on the first concept map and the second concept map, related words that indicate the concept common to the search word for the first user and the second user belonging to the first group are extracted. Since a search is performed based on the extracted related words and search words, even if the first user or the second user has never purchased a product or used a service, etc., the first group can It is possible to recommend the most suitable product or service for the customer.

Furthermore, as described above, the difference v− A general word whose v' is greater than or equal to the first threshold is adopted as a related word. As a result, it is possible to extract related words indicating concepts common only to the first user and the second user belonging to the first group, and to increase the accuracy of extracting related words.

(Embodiment 2)
[2-1. Configuration of Recommended Concept Calculation Unit]
The configuration of the recommended concept calculation unit 12A of the search system 2A according to Embodiment 2 will be described with reference to FIGS. 19 and 20. FIG. FIG. 19 is a block diagram showing the configuration of a search system 2A according to Embodiment 2. As shown in FIG. FIG. 20 is a diagram showing an example of calculation of the conceptual average distance by the conceptual average distance calculator 90 according to the second embodiment. In each of the following embodiments, the same reference numerals are given to the same constituent elements as in the first embodiment, and the description thereof will be omitted.

As shown in FIG. 19, in a search system 2A according to the second embodiment, the configuration of a recommended concept calculation unit 12A is different from that in the first embodiment. Specifically, the recommended concept calculation unit 12A has an average concept distance calculation unit 90 in addition to the components described in the first embodiment.

The concept average distance calculation unit 90 calculates the concept average distance based on the first concept map and the second concept map received by the recommendation target terminal concept map reception unit 30, and applies the calculated concept average distance to the overall common concept. Output to the exclusion unit 38A. Specifically, the conceptual average distance calculation unit 90 calculates the conceptual average distance D _aveij based on Equation 5 below.

In Equation 5 above, D _aveij is the conceptual average distance of row i and column j, d _xij is the semantic distance of row i and column j of the first concept map, and d _yij is row i of the second concept map. It is the semantic distance of the j column. Here, the concept average distance is a value based on the sum of the semantic distance of the first concept map and the semantic distance of the second concept map. The larger the concept average distance, the semantic distance between the general word in row i and the general word in column j of the first concept map, and the general word in row i and the general word in column j of the second concept map. At least one of the semantic distances between and is large. That is, the greater the conceptual average distance, the greater the semantic distance between the search word and the general word for at least one of the first user and the second user. As shown in FIG. 20, the conceptual average distance calculator 90 calculates the semantic distance of the first concept map shown in FIG. 4A and the semantic distance of the second concept map shown in FIG. Compute the conceptual mean distance between In FIG. 20, a line surrounded by a dashed frame line 92 represents the conceptual average distance of each of a plurality of general words with respect to the search word "home".

[2-2. Operation of Global Common Concept Exclusion Part]
Next, referring to FIG. 21, the operation of the global common concept exclusion unit 38A will be described. FIG. 21 is a flow chart showing the operation flow of the global common concept exclusion unit 38A according to the second embodiment. In the flowchart of FIG. 21, the same step numbers are assigned to the same processes as those of the flowchart of FIG. 11 described above, and the description thereof will be omitted.

First, steps S301 to S306 are executed as in the first embodiment. In step S306, if the difference v−v′ is equal to or greater than the first threshold (YES in S306), the overall common concept exclusion unit 38A determines that the concept average distance is equal to or less than the second threshold (for example, “10”). It is determined whether or not there is (S401).

If the average concept distance is equal to or less than the second threshold (YES in S401), the global common concept exclusion unit 38A adopts the extracted general words as related words (S307). On the other hand, if the average concept distance exceeds the second threshold (NO in S401), the global common concept exclusion unit 38A does not adopt the extracted general word as a related word (S308).

[2-3. effect]
As described above, when the concept average distance exceeds the second threshold, the global common concept exclusion unit 38A does not adopt the extracted common word as a related word. As a result, it is possible to exclude from the related words general words that indicate a concept common to the search word for the first user and the second user, but are semantically distant from the search word. As a result, a more optimal product or service can be recommended to the first group.

(Embodiment 3)
The configuration of a search system 2B according to Embodiment 3 will be described with reference to FIG. FIG. 22 is a block diagram showing the configuration of a search system 2B according to Embodiment 3. As shown in FIG.

As shown in FIG. 22, in the search system 2B according to Embodiment 3, the recommended concept calculation section 12 and the recommended service provision section 14 are installed in the same relay device 94 (an example of a search device). Even with such a configuration, the same effects as in the first embodiment can be obtained.

(Embodiment 4)
A configuration of a search system 2C according to Embodiment 4 will be described with reference to FIG. FIG. 23 is a block diagram showing the configuration of a search system 2C according to the fourth embodiment.

As shown in FIG. 23, in the search system 2C according to the fourth embodiment, the recommended concept calculation unit 12 is installed in each of the recommended user terminals 16Ca and 16Cb (an example of a search device). The recommended concept transmitted from the recommendation target user terminal 16Ca is received by the recommendation service providing unit 14. FIG. Even with such a configuration, the same effects as in the first embodiment can be obtained.

(Other modifications)
As described above, the word expansion method and the like according to one or more aspects have been described based on the first to fourth embodiments, but the present disclosure is not limited to the first to fourth embodiments. As long as it does not deviate from the spirit of the present disclosure, various modifications that a person skilled in the art can think of are applied to the embodiment, and a form constructed by combining the components of different embodiments is also within the scope of one or more aspects. may be included in

In each of the above embodiments, the overall common concept extraction unit 36 used the first to fourth concept maps when extracting related words. A concept dictionary such as Note that wordnet is described, for example, in "George A. Miller" WordNet: A Lexical Database for
English", Communications of the ACM, Volume 38 Issue 11, Nov. 1995".

In each of the above-described embodiments, the overall common concept extraction unit 36 used the first to fourth concept maps when extracting related words, but the fourth concept map was omitted, and at least the first to third concept maps were used. You may use the concept map of

In each of the above-described embodiments, the database 44 of the recommendation service providing unit 14 stores information about residential properties, but is not limited to this, and may store various information about travel destinations, accommodations, restaurants, etc. good.

In each of the above embodiments, the semantic distance in each of the first to fourth concept maps is represented by an integer value of "0" to "10", but is not limited to this. It may be represented by a decimal value such as "1.0".

In each of the above embodiments, the smaller the semantic distance value, the closer the semantic distance between each of a plurality of general words. A larger value may represent a closer semantic distance between each of a plurality of common words.

In each of the above-described embodiments, each component may be configured by dedicated hardware, or realized by executing a software program suitable for each component. Each component may be realized by reading and executing a software program recorded in a recording medium such as a hard disk or a semiconductor memory by a program execution unit such as a CPU or processor.

Also, part or all of the functions of the search device according to each of the above embodiments may be implemented by a processor such as a CPU executing a program.

A part or all of the components constituting each device described above may be configured by an IC card or a single module that can be attached to and detached from each device. The IC card or module is a computer system composed of a microprocessor, ROM, RAM and the like. The IC card or the module may include the super multifunctional LSI. The IC card or the module achieves its function by the microprocessor operating according to the computer program. This IC card or this module may have tamper resistance.

The present disclosure may be the method shown above. Moreover, it may be a computer program for realizing these methods by a computer, or it may be a digital signal composed of the computer program. In addition, the present disclosure includes a computer-readable recording medium for the computer program or the digital signal, such as a flexible disk, hard disk, CD-ROM, MO, DVD, DVD-ROM, DVD-RAM, BD (Blu-ray (registered (trademark) Disc), a semiconductor memory, or the like. Alternatively, the digital signal recorded on these recording media may be used. Further, according to the present disclosure, the computer program or the digital signal may be transmitted via an electric communication line, a wireless or wired communication line, a network represented by the Internet, data broadcasting, or the like. The present disclosure may also be a computer system comprising a microprocessor and memory, the memory storing the computer program, and the microprocessor operating according to the computer program. Also, by recording the program or the digital signal on the recording medium and transferring it, or by transferring the program or the digital signal via the network, etc., it is implemented by another independent computer system It is good as

A search method according to the present disclosure is useful for a search system or the like for recommending products or services to a specific group.

2, 2A, 2B, 2C search system 4 group 6 user 8 voice terminal 10 information terminal 12, 12A recommendation concept calculation section 14 recommendation service provision section 16a, 16b, 16Ca, 16Cb recommended user terminal 18a, 18b non-recommended user terminal 20 Internet 22 Search word transmitting unit 24 Concept map holding unit 26 Recommendation result receiving unit 28 Search word receiving unit 30 Recommended terminal concept map receiving unit 32 Non-recommended terminal concept map receiving unit 34 Group common concept extracting unit 36 Overall common concept extracting Units 38, 38A Overall common concept exclusion unit 40 Recommended concept transmission unit 42 Recommended concept reception unit 44 Database 46 Recommendation request unit 48 Recommendation content generation unit 50 Recommendation result transmission unit 52, 54, 92 Frame lines 56, 62 Group common concept 58, 60 Overall common concept 64 Related words 66 Search query 68 Property database 70 Display units 72, 74, 86, 88 Display areas 76, 78 Option buttons 80 Recommendation result display screens 82, 84 Recommendation result 90 Concept average distance calculation unit 94 Relay device

Claims (15)

A search method performed by a processor, comprising:
The processor
(a) obtaining a search word entered by a first user belonging to a first group or a second user belonging to said first group ;
(b) a first concept map, a second concept map and a third concept map comprising a plurality of words and a semantic distance between each of said plurality of words, said first group; a first concept map specific to the first user belonging to the first group, a second concept map specific to the second user belonging to the first group, and a second group different from the first group obtaining from memory a third concept map specific to a third user belonging to
(c) obtaining a first relevance map including a relevance indicating a closeness between the semantic distance included in the first concept map and the semantic distance included in the second concept map; ,
(d) the closeness of the semantic distances included in the first concept map, the semantic distances included in the second concept map, and the semantic distances included in the third concept map; obtain a second relevance map containing relevances indicating
(e) a difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map among the plurality of words; is greater than or equal to the first threshold, extracting at least one word as a related word;
(f) outputting a result of searching based on the search word and the related word to at least one of the first user and the second user ; In (e) above, for the at least one word whose difference is equal to or greater than the first threshold, the semantic distance from the search word included in the first concept map and the semantic distance to the search word in the second concept map 2. The search method according to claim 1, wherein when a value based on the sum of said semantic distances to said search words included is equal to or less than a second threshold, said related words are extracted. 3. The method according to claim 1 or 2, wherein in (f), a result of searching based on the search word and the related word is displayed to the first user or the second user belonging to the first group. Described search method. wherein in (f), the search results based on the search words and the related words are displayed in descending order of the relevance between the search words and the related words included in the first relevance map. The search method described in 3. 5. The search method according to claim 4, wherein in (f), a result of searching based on the search word and the related word is displayed at a higher rank than contents highly evaluated by other users. The retrieval method according to any one of claims 1 to 5, wherein each of said first concept map, said second concept map and said third concept map is generated based on results of brain measurement. Each of the first conceptual map, the second conceptual map and the third conceptual map is generated based on the results of brain measurement using fMRI (Functional Magnetic Resonance Imaging). retrieval method. A retrieval device comprising a processor and a memory,
The processor
(a) obtaining a search word entered by a first user belonging to a first group or a second user belonging to said first group ;
(b) a first concept map, a second concept map and a third concept map comprising a plurality of words and a semantic distance between each of said plurality of words, said first group; a first concept map specific to the first user belonging to the first group, a second concept map specific to the second user belonging to the first group, and a second group different from the first group obtaining from said memory a third concept map specific to a third user belonging to
(c) obtaining a first relevance map including a relevance indicating a closeness between the semantic distance included in the first concept map and the semantic distance included in the second concept map; ,
(d) the closeness of the semantic distances included in the first concept map, the semantic distances included in the second concept map, and the semantic distances included in the third concept map; obtain a second relevance map containing relevances indicating
(e) a difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map among the plurality of words; is greater than or equal to the first threshold, extracting at least one word as a related word;
(f) A search device that outputs a search word to at least one of the first user and the second user based on the search word and the related word. In (e) above, for the at least one word whose difference is equal to or greater than the first threshold, the semantic distance from the search word included in the first concept map and the semantic distance to the search word in the second concept map 9. The search device according to claim 8, wherein the search word is extracted as the related word when a value based on the sum of the semantic distance with respect to the included search word is equal to or less than a second threshold. The processor further
(g) displaying a result of searching based on the search word and the related word to the first user or the second user belonging to the first group; Device. wherein in (g), the search results based on the search words and the related words are displayed in descending order of the relevance between the search words and the related words included in the first relevance map. 11. The search device according to 10. 12. The search device according to claim 11, wherein in (g), a result of searching based on the search word and the related word is displayed at a higher rank than content highly evaluated by other users. The retrieval device according to any one of claims 8 to 12, wherein each of said first concept map, said second concept map and said third concept map is generated based on brain measurement results. 14. The retrieval device according to claim 13, wherein each of said first concept map, said second concept map and said third concept map is generated based on results of brain measurement using fMRI. A program that causes a computer to perform a search method, comprising:
to the computer;
(a) obtaining a search word entered by a first user belonging to a first group or a second user belonging to said first group ;
(b) a first concept map, a second concept map and a third concept map comprising a plurality of words and a semantic distance between each of said plurality of words, said first group; a first concept map specific to the first user belonging to the first group, a second concept map specific to the second user belonging to the first group, and a second group different from the first group obtaining from memory a third concept map specific to a third user belonging to
(c) obtaining a first relevance map including a relevance indicating a closeness between the semantic distance included in the first concept map and the semantic distance included in the second concept map; ,
(d) the closeness of the semantic distances included in the first concept map, the semantic distances included in the second concept map, and the semantic distances included in the third concept map; obtain a second relevance map containing relevances indicating
(e) a difference between the degree of relevance for the search word included in the first degree-of-relevance map and the degree of relevance for the search word included in the second degree-of-relevance map among the plurality of words; is greater than or equal to the first threshold, extracting at least one word as a related word;
(f) outputting a search result based on the search word and the related word to at least one of the first user and the second user ;

Images